See discussions, stats, and author profiles for this publication at: [Link]

net/publication/318695199

Implementation of hybrid Kanban-CONWIP system: A case study

Article  in  Journal of Manufacturing Technology Management · July 2017

DOI: 10.1108/JMTM-03-2016-0043

CITATIONS READS

14 662

6 authors, including:

Bruno Sereno Mauro Sampaio

Faculdade de Tecnologia de São Bernardo do Campo Centro Universitario FEI
2 PUBLICATIONS   15 CITATIONS    39 PUBLICATIONS   152 CITATIONS   

SEE PROFILE SEE PROFILE

Alexandre Augusto Massote Jairo Simoes

University Center of FEI Universidade Presbiteriana Mackenzie
19 PUBLICATIONS   39 CITATIONS    2 PUBLICATIONS   14 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Supply Chain View project

Engineering Education View project

All content following this page was uploaded by Mauro Sampaio on 27 December 2019.

The user has requested enhancement of the downloaded file.

 Journal of Manufacturing Technology Management
Implementation of hybrid Kanban-CONWIP system: a case study
Dênis Gustavo Leonardo, Bruno Sereno, Daniel Sant Anna da Silva, Mauro Sampaio, Alexandre
Augusto Massote, Jairo Celso Simões,
Article information:
To cite this document:
Dênis Gustavo Leonardo, Bruno Sereno, Daniel Sant Anna da Silva, Mauro Sampaio, Alexandre
Augusto Massote, Jairo Celso Simões, (2017) "Implementation of hybrid Kanban-CONWIP system: a
case study", Journal of Manufacturing Technology Management, Vol. 28 Issue: 6, pp.714-736, https://
[Link]/10.1108/JMTM-03-2016-0043
Permanent link to this document:
[Link]
Downloaded on: 07 January 2018, At: 18:06 (PT)
Downloaded by FEI At 18:06 07 January 2018 (PT)

References: this document contains references to 24 other documents.

To copy this document: permissions@[Link]
The fulltext of this document has been downloaded 199 times since 2017*
Users who downloaded this article also downloaded:
(2017),"Lean manufacturing practices in Indian manufacturing SMEs and their effect on sustainability
performance", Journal of Manufacturing Technology Management, Vol. 28 Iss 6 pp. 772-793 <a
href="[Link]
(2017),"Generic decision support system to leverage supply chain performance (GLE) for SMEs
in Thailand", Journal of Manufacturing Technology Management, Vol. 28 Iss 6 pp. 737-748 <a
href="[Link]

Access to this document was granted through an Emerald subscription provided by emerald-
srm:571304 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald
for Authors service information about how to choose which publication to write for and submission
guidelines are available for all. Please visit [Link]/authors for more information.
About Emerald [Link]
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as
well as providing an extensive range of online products and additional customer resources and
services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the
Committee on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for
digital archive preservation.

*Related content and download information correct at time of download.

 The current issue and full text archive of this journal is available on Emerald Insight at:
[Link]/[Link]

JMTM
28,6 Implementation of hybrid
Kanban-CONWIP system:
a case study
714 Dênis Gustavo Leonardo, Bruno Sereno, Daniel Sant Anna da Silva,
Received 30 March 2016
Mauro Sampaio and Alexandre Augusto Massote
Revised 26 September 2016 Industrial Engineering Department, Centro Universitario da FEI,
1 April 2017
17 May 2017
Sao Bernardo do Campo, Brazil, and
Accepted 26 May 2017 Jairo Celso Simões
Electrical Engineering Department,
Universidade Presbiteriana Mackenzie, São Paulo, Brazil

Abstract
Downloaded by FEI At 18:06 07 January 2018 (PT)

Purpose – Shop floor control systems are generally major points of discussion in production planning and
control literature. The purpose of this paper is to investigate how lean production control principles can be
used in a make-to-order (MTO) job shop, where the volume is typically low and there is high variety.
This paper examines the procedures involved in implementing a constant work-in-process (CONWIP)/Kanban
hybrid system in the shop floor environment and also provides insights and guidelines on the implementation
of a hybrid system in a high-variety/low-volume environment.
Design/methodology/approach – The authors review literature on Kanban, CONWIP, and CONWIP/
Kanban hybrid systems to analyze how lean production control principles can be used in a MTO job shop.
The second part focuses on the process of implementation. Using a case study of a manufacturer of
electromechanical components for valve monitoring and controls, the paper describes how the operation is
transformed by for more efficient shop floor control systems. Real experiments are used to compare pre- and
post-improvement performance.
Findings – The study shows that the proposed hybrid Kanban-CONWIP system reduced the cycle time and
achieved an increase of 38 percent in inventory turnover. The empirical results from this pilot study provide
useful managerial insights for a benchmarking analysis of the actions to be taken into consideration by
companies that have similar manufacturing systems.
Research limitations/implications – The statistic generalization of the results is impossible due to the
use of a single case method of study.
Originality/value – This paper provides insights and guidelines on the implementation of a hybrid system
in a high-variety/low-volume environment. The literature on real applications of hybrid CONWIP/Kanban by
case study is limited.
Keywords Lean production, Production control, Kanban
Paper type Case study

1. Introduction
Several authors have already studied and discussed production planning and control
in the literature (Fernandes and Godinho Filho, 2010; Prakash and Chin, 2015;
Framinan et al., 2003). We are familiar with several pull systems such as production
control mechanism (PCM), namely Kanban, constant work-in-process (CONWIP), hybrid
CONWIP/Kanban, among others (Prakash and Chin, 2015). A PCM aims at reaching high
customer service level without increasing inventories; however, the key issue today is to find
the one which has the best performance in a specific environment (Lage and Godinho Filho,
2008; Farnoush and Wiktorsson, 2013).
Journal of Manufacturing
Technology Management
Vol. 28 No. 6, 2017
pp. 714-736 The authors would like to thank the anonymous reviewers for their helpful comments that helped
© Emerald Publishing Limited improve the presentation and content of this paper. The authors also would like to acknowledge the
1741-038X
DOI 10.1108/JMTM-03-2016-0043 financial support from the FEI University.
 The most popular is Kanban (Ohno, 1988; Shingo and Dillon, 1989), wherein a product is Hybrid
produced by a workstation according to its immediate downstream needs; however, Kanban Kanban-
is slow to respond to demand changes. Another widely known system is CONWIP CONWIP
(Spearman et al., 1990) production control system. In practice, this means that when certain
parts are required at the end of a line, card(s) is/are passed to the beginning of the line to system
fulfill that need. This procedure limits WIP inventories within a line. Spearman et al. (1990)
conclude that CONWIP is more responsive to changes in customer demand. 715
The combination of these systems gave rise to a production system called the hybrid
CONWIP/Kanban (Bonvik et al., 1997). Nevertheless, we still know little about the possible
gains related to the adoption of these methods, mainly when it comes to small batch
productions with high customization and uncertain demand. Literature on CONWIP system
modeling and simulation is rich and can be used to build and enhance theoretical
understanding. Nevertheless, literature on the real applications of hybrid CONWIP/Kanban
by case study is limited (Framinan et al., 2003; Prakash and Chin, 2014).
This research aims at understanding, describing, and investigating the following
production control methods: Kanban, CONWIP, and the hybrid CONWIP/Kanban. It will
identify the main difficulties and bottlenecks companies face to implement these pull
Downloaded by FEI At 18:06 07 January 2018 (PT)

systems, the gain in productivity, and mainly the increase in the level of service of a
company with small batch production, which has high customization and uncertain
demand. It will demonstrate that these aspects are relevant to the performance of control
systems. We have developed all concepts related to Kanban, CONWIP, and the CONWIP/
Kanban hybrid method in order to explore these pull systems and discover which of them
best fits the company under study. Guidelines for the reproduction of the system for similar
production environments are also provided.
In short, we intend to answer the following research questions:
RQ1. What were the project’s results and what were the key variables that have
determined them?
RQ2. When is the hybrid CONWIP/Kanban more appropriate in productive systems?
RQ3. What are the results obtained with the application of the hybrid CONWIP/Kanban
in a low variety/low volume environment?
By analyzing a case at a manufacturer of electromechanic accessories to manage and
monitor valves, we will attempt to provide insights and guidelines on the implementation of
a hybrid system in a low-variety/low-volume environment. Alternative procedures for
releasing materials into the process are evaluated in order to determine which of them offers
the best advantage to managers.
The remainder of this paper is organized as follows. In Section 2, we summarize the
literature related to pull systems. This is followed by a presentation of the method.
In following section, we present in details: the case study, the problem, and implementation
evolution. In the concluding section, we draw and incorporate suggestions to further
research on pull systems.

2. Theoretical background
Leaders of industries face the typical problem of improving the performance of processes in
manufacturing industries (Fernandes and Godinho Filho, 2010). In the 1980s, the system
boosted by the Japanese industry, named Kanban, became popular all over the world. In the
1990s, a team of researchers from the Massachusetts Institute of Technology (MIT) declared
it to be the most competitive production model among the ones adopted by automakers
throughout the world.
 JMTM However, researchers observed over the years that Kanban did not behave satisfactorily
28,6 when subjected to a variable mix of products with varying levels of demand. In this case,
it resulted in high inventory levels, a reduction in production, an increase in storage costs,
and unsatisfactory levels of service. So Kanban is only suitable for a high-volume
production environment with relatively few part types (Bonvik et al., 1997).
Ten years later, the CONWIP method was proposed (Spearman et al., 1990), which,
716 according to the authors, solved the problems found in Kanban. Similarly, the hybrid
CONWIP/Kanban (Bonvik et al., 1997) was proposed based on the results of the experience
with the application of these two methods separately. It aimed at incorporating their
respective positive aspects, whose combination would generate a significant increase in the
level of customer service.

2.1 Kanban
Kanban is a Japanese word that means visual record and refers to a manufacturing control
system developed and used in Japan. The Kanban cards are used for information
communication and inventory control. Its main function is to control inventory levels,
keeping them as low as possible without compromising production. Kanban has control
Downloaded by FEI At 18:06 07 January 2018 (PT)

loops that connect each production stage with its immediate predecessor, as Figure 1 shows.
When the work-in-process (WIP) of the following stage is low, Kanban allows production to
start at the previous stage. According to Boonlertvanich (2005), we can find several versions
of Kanban control, such as single card, two cards, etc. The comparison of the several
Kanban systems is present in the work of Muckstadt and Tayur (1995).
A Kanban system is very efficient in an ideal environment: low process and demand
variability, few breakdowns, and so on. The main drawback of the Kanban system is that
the system, especially in the upstream stages, may not respond quickly enough to the
changes in demand (Sharma and Agrawal, 2009).

2.2 CONWIP
CONWIP is a shop floor control procedure presented in the work of Spearman et al. (1990),
which sets a limit on the total WIP in the entire system in order to reduce costs and lead time.
As Figure 2 shows, the CONWIP system has a single loop from the final to the initial

Request Request Request

Raw First Last

materials stage stage

Inventory Inventory Finished goods

Figure 1. inventory
Kanban system
Source: Adapted from Gaury et al. (2009)

Request

Raw First Last

materials stage stage

Inventory Inventory Finished goods

Figure 2. inventory
CONWIP system
Source: Adapted from Gaury et al. (2009)
 production stage. The system will release jobs into the system whenever another job Hybrid
completes processing at last workstation. As CONWIP feedbacks the message of demands Kanban-
immediately to the first work center, some researchers called it a single-stage Kanban system. CONWIP
In effect, the CONWIP control system can be considered as a pull system at the end of the line,
or a push system from the beginning of the line toward the end (Boonlertvanich, 2005). system
CONWIP is applicable to production environments that produce many kinds of products.
The CONWIP system seeks to control the plant at a constant level of WIP although it does 717
not attempt to control the location of the WIP within the system.

2.3 Hybrid Kanban/CONWIP

According to Bonvik et al. (1997), a disadvantage of CONWIP is that inventory levels inside
the system are not controlled individually, for example, high inventories can appear in front
of slow machines. Therefore, the authors proposed a new control strategy called the hybrid
CONWIP/Kanban, in which the CONWIP control is completed with secondary cells
controlled by the Kanban, as seen in Figure 3. The last stage does not need a Kanban
control, since any part that has progressed so far will replace a delivered finished good part
(CONWIP control principle).
Downloaded by FEI At 18:06 07 January 2018 (PT)

For Gaury et al. (2000), the idea is to combine the advantages of CONWIP (high throughput
with a low overall WIP level) with those of Kanban (control of inventory levels at each stage).
Bonvik et al. (1997) show that the advantage of hybrid Kanban/CONWIP over Kanban
grows with the length of the process, the degree of process variability, and the service-level
target. CONWIP systems handle a mix of parts having different bottlenecks with more ease
than Kanban systems.

2.4 Real applications of CONWIP

There are a number of existing studies about comparing Kanban, CONWIP, and hybrid
systems. The majority of the existing literature used a conceptual model or a simplified
problem to illustrate CONWIP/pull control strategies effectiveness considering various
performance parameters. However, literature on the implementation and empirical
analysis of case studies is limited. The applicability of CONWIP/pull control strategies in
solving a practical size problem is not well-documented in the literature (Yang et al., 2007).
Table I shows the main case studies published about CONWIP. We will briefly synthesize
each of them, highlighting their contributions and the main points related to the theme of
this research.
Huang et al. (1998) were the first authors to introduce a simulation study on the
application of CONWIP in a semi-continuous manufacturing environment, a cold rolling
plant. The simulation study compares the CONWIP system and the original control system
for the four situations in a cold rolling plant. The simulation’s results show that the
CONWIP production control system is very efficient for the production and inventory
control of semi-continuous manufacturing. It can greatly reduce the WIP, decrease the

C
K1 K2 K c–1

Raw First Last

materials stage stage
Inventory Inventory Inventory Finished goods Figure 3.
inventory Hybrid Kanban/
CONWIP
Source: Adapted from Gaury et al. (2009)
 JMTM average inventory and average inventory costs, and guarantee a higher throughput rate and
28,6 facility utilization.
Bonvik et al. (1997) studied the performance of the Kanban, minimal blocking, base
inventory, CONWIP, and hybrid Kanban-CONWIP control policies in a four-machine
tandem production line making parts for an automobile assembly line. The authors found
that the best parameter choices for the hybrid policy decrease inventories by 10-20 percent
718 over the best Kanban policy while maintaining the same service levels.
Gilland (2002) analyzed the problem of managing the flow of material into a
manufacturing cell in order to maximize the output rate for a given level of WIP inventory.
The author demonstrated that when one or more bottlenecks are present, a policy called
drum-buffer-rope outperforms other release rules, even the CONWIP system.
Yang et al. (2007) addressed an evolutionary-simulation optimization approach in solving
a multi-CONWIP pull strategy problem. The authors state that a simulation-optimization
approaching could be a barrier to solve a practical problem because a real problem required
a significant amount of modeling time and experiences to build a simulation model.
Slomp et al. (2009) presented the development of a lean production control system for a
high-variety/low-volume production unit. The main contribution is the illustration that lean
Downloaded by FEI At 18:06 07 January 2018 (PT)

control principles, such as leveling, pull and takt time can be used in such a context. The lean
production control system has led to short and reliable flow times on the shop floor.
Farnoush and Wiktorsson (2013) presented a study on the applicability of four different PCM
in an automotive production company with high variety of products and divergent lines. Their
study shows that m-CONWIP is recommended in this situation, as it leads to more reduction in
the WIP amount and throughput time, while the throughput value does not decrease.
Although many practical cases prove that CONWIP systems are effective, only Prakash and
Chin (2014) have investigated the implementation of a modified Kanban and hybrid system.
This case study proves that pull systems can be successfully implemented in production
environments that do not conform to the typical prerequisites of the Kanban system.
Ni and Wang (2015) assisted manufacturers in achieving real-time ordering and
obtaining integrated optimization of shop floor production planning and scheduling for
mixed flow production systems.
With this literature revision, we can conclude that CONWIP and the hybrid CONWIP/
Kanban systems are of the interest of researchers, but most of the studies are still focused on
simulation, as Table I shows. The occurrence is rare in real life.

3. Methodology
The purpose of this research is to investigate an emerging phenomenon, the adoption of the
practice of Hybrid CONWIP/Kanban in the Brazilian business environment, an event over
which the researcher has no control. In this context, the methodology suggested by several

References System characteristics Simulation Test pilot CONWIP Hybrid method

Huang et al. (1998) Cold rolling plant | |

Bonvik et al. (1997) Automobile assembly line | | |
Gilland (2002) “Intel” microprocessors factory | |
Yang et al. (2007) Integrated circuit | |
Table I. Slomp et al. (2009) Conductor components factory | | |
Research works Farnoush and Automotive production
that address Wiktorsson (2013) company | |
real application Prakash and Chin (2014) Composite subassemblies | |
of CONWIP Ni and Wang (2015) Semiconductor assembly | |
 authors (Ellram, 1996; Yin, 2013; Eisenhardt, 1989) is the case study. The available Hybrid
knowledge on the phenomenon of interest is scarce, observed by the low index of Kanban-
publications on the topic. In this case, an exploratory research is essential to develop and CONWIP
provide new ideas and discoveries.
According to Yin (2013), we can find two types of case studies: a single case study, system
in which the unit of analysis is only one case; and multi-case studies, in which several cases
are analyzed in order to allow comparisons. The author also classifies the single case study 719
as holistic or incorporated. The holistic type has a unit of analysis considered in a specific
context. On the other hand, the incorporated type has more than one unit of analysis for
each case, i.e. it has sub-units of analysis.
Yin (2013) establishes another classification of the case study regarding the
research objective: descriptive (describing the phenomenon within its context),
exploratory (dealing with little known problems, aiming at defining hypotheses or
propositions for future researches), or explanatory (intending to explain the relations of
cause and effect from a theory).
Therefore, according to Yin’s parameters, this study is presented as single case, holistic,
exploratory, mainly due to the rare nature of finding cases to apply the hybrid CONWIP/
Downloaded by FEI At 18:06 07 January 2018 (PT)

Kanban in the market.

In the case study methodology, it does not seem to make sense to establish a formal a
priori hypothesis. However, according to Yin (2013), the establishment of study propositions
helps the researcher to guide the task of gathering information:
P1. The choice of the production control method is directly related to the characteristics
found in each productive process, such as size of the batch, degree of demand forecast,
level of repetition, diversification (variety of different products), and distinction
(variety of models), as defined in the studies by MacCarthy and Fernandes (2000) and
Fernandes and Godinho Filho (2010).
P2. The CONWIP/Kanban hybrid method is more productive in a high-variety/
low-volume environment.
P3. The results achieved via the hybrid CONWIP/Kanban are superior to the results
achieved through methods of Kanban and CONWIP in the production environments
of small batches with customized products.
One of the problems, which are most commonly related to the case study, is external
validity, as it is impossible to generalize from single case studies. The objective of
this study was not to carry out statistical generalizations of the results, but only
analytical generalizations.
According to Yin (2013), the validity and reliability of a study are also very dependent on
techniques of collection and treatment of data used, and the adaptation of these techniques
to the research objectives. If these techniques are selected and properly applied, with the
procedures explained, the problems will be reduced.

3.1 Unit of analysis

The unit of analysis for the present case study was a multinational company from the
industrial automation sector, which adopts the hybrid CONWIP/Kanban. The level of analysis
considered was the opinion of its decision makers: executives, directors, and/or managers.

3.2 Case selection

We selected a company with a very peculiar characteristic – the production of small batches
with high customization. The objective of this analysis was to understand the
implementation process of the production control methods (Kanban, CONWIP, hybrid
 JMTM CONWIP/Kanban), analyze the generating needs, highlight the problems faced, and finally
28,6 discuss the results obtained with the implementation of their system for coordinating
purchase and production orders.

3.3 Data collection protocol

Several authors (Yin, 2013; Eisenhardt, 1989) state that the collection of qualitative data
720 must be based on multiple sources of evidence, but always converging to the same set of
facts and discoveries. Qualitative data are considered evidence of the connection between
data and propositions. We have carried out this present case study including in-depth
interviews with the company’s personnel, as Table II shows. Eight meetings were held
which lasted approximately three hours, with leaders of the implementation process of the
production control methods in the company.

3.4 Interviewee
The pilot project was led by the operational manager of the company’s unit in Brazil
who is also the lean engineer of the company. The main source of data was the interviews
Downloaded by FEI At 18:06 07 January 2018 (PT)

carried out with the production and operations managers. We have obtained information
regarding records, operational and managerial, performance indicators, market
positioning, among others.
Yin (2013) warns researchers that it is important to have an analytical strategy to collect
evidence from a case study before starting any field studies. The analysis structure of
the case study was: first, we carried out a simple narrative of the case at issue; and second,
we dedicated a section to show the historical results and posterior to the implementation
of the method. A matrix of the important variables was structured during the interviews
with the company’s executive and personnel.

4. Results
4.1 Company and project background
The case study was a subsidiary company of a North American group, manufacturer of
hi-tech electromechanical products, including innovative solutions to manage and monitor
valves. The company studied is a global firm with production, sales, and facilities in Brazil,
the USA, and the UK.
The company only represented the brand in Brazil until 1990. Its operation started
in 1991, as a subsidiary in the city of São Paulo, operating exclusively as an importer.
In 1997, it started assembling completely knock-down sets imported from other global
units. In 2010, it started nationalizing its products to the Brazilian market, from
development to manufacture.
At the end of 2014, the company had 180 employees divided into the following areas: five
employees in management, 15 in sales, 28 in administration, and 132 in operation, working
with four production lines divided into four cells each, with local revenue of approximately
20 million dollars per year. The company worked with the following types of production:
make to inventory and make to order.

Department Position

Administrative Operational manager

Table II. Operations Process engineer
List of interviewees Quality engineer
 4.2 Studied period Hybrid
This study focuses on the events from 2012 to 2014, when a series of changes happened at Kanban-
the company. We intend to demonstrate these changes through the result analysis of the CONWIP
three main years of this period – 2012, 2013, and 2014.
system
4.3 The case study problem
A steady increase in product demand was verified, followed by an increase in production, 721
when the company started to move manufacturing back from China to a factory in São
Paulo, which not long ago had been expected to close. As a result, the mix of products
increased and demands became even more varied, without any forecast possibility.
In 2011, the company marketed a total of SKU39,500, equivalent to 63 distinct families.
Figure 4 indicates a Pareto chart of the families on demand within the studied period.
The number of families marketed is on the x-axis on a logarithmic scale, and the quantities
of items per family are indicated on the y-axis.
The families were subdivided into three main groups: A, B, and C.
(1) Group A – hatched columns: families with demand over 180 pieces per month. This
Downloaded by FEI At 18:06 07 January 2018 (PT)

group represents 6 percent of the variations, 64 percent of the items produced;

however, only 10 percent of sales revenue.
(2) Group B – columns without filling: families with demand under 180 pieces and over
12 pieces per month. This group represents 16 percent of the variations, 29 percent
of the items produced, and it corresponds to 56 percent of sales revenue.
(3) Group C – dot-filled columns: families with demand less than 12 pieces per month.
This group represents the largest percentage of variations, totaling 78 percent, with
only 7.5 percent of the produced items and a total of 34 percent of sales revenue.
In Figure 5, we can verify how the three pieces of information in each group relate: quantity
of supplied items, turnover, and number of distinct families.

10,000
4 family of products with demand >180 units/months
10% turnover
6% product families
64% production volume

1,000 10 family of products with 180 > demand >12 units/months

56% turnover
16% product families
29% production volume
Demand

49 family of products with demand

100 <10 units/months
34% turnover
78% product families
7% production volume

10

1
Figure 4.
Pareto chart of
K-SWITCH
EC2020/2080
2004
9479
2007
BF9358
7644
EP000
7544
MCAL
7379
IP10
2085
711
9479 C/ SOL
AVID
1040
BF366 C/ SOL
BF765
3479
7645
7699
40-A
D7644
4210
8749
4230
2869
9468
5235
7179
4330

the families
Product
by logarithmic
family
demand – reference
data: January 2011-
December 2011
Source: The Company
 JMTM 90%

28,6 80%

70%

60%

50%
722 40%

30%

20%

10%

0%
Group A Group B Group C
Family 6% 16% 78%
Production 64% 29% 7%
Figure 5. Turnover 10% 56% 34%
Groups A, B, and C
Source: The Company
Downloaded by FEI At 18:06 07 January 2018 (PT)

We can verify that the three groups of products studied have their importance. They can
occupy production capacity with high volumes (Group A), represent a large sale revenue
(Group B), or meet the needs of the market through the quantity of families (Group C).
In order to illustrate the variation and lack of predictability of demand which is intrinsic
to this type of market and inherent to the family of products at issue, Figure 6 refers to the
behavior of three of these families which belong to the Groups A, B, and C, during the period
of approximately three years.
From the analysis of the graph in Figure 6, we can observe that regular seasonality is not
present in the volumes consumed, which hinders programming production based on
forecast, given that the demand is unpredictable.

2011 2012 2013 2014

2,100

1,800

1,500

1,200

900

600

300

Figure 6. 0
J A S O N D J F M A M J J A S O N D J F M A M J J A S O N D J F
Demand for finished
products – Group A, Group A Group B Group C
B, and C
Source: The Company
 The company’s problem in this scenario was how to program production, taking into Hybrid
account this variable mix of products and low sales predictability. In order to solve this Kanban-
issue, the company’s managers implemented a series of processes/methodologies, which will CONWIP
be approached henceforth.
system
4.4 Processes and implementation
As a result of the nationalization of products and processes and the increase in volume, the 723
need to adapt the productive process to some production control method arises.
Following the matrix guidelines, the company decided to implement production control
through manufacturing resource planning (MRP), programming production with analysis of
demand forecast based on the records. This process had a maturation period of almost
two years before being implemented.
As a result of this adaptation, the production line was characterized as illustrated
in Figure 7.
The MRP interacts with the master production schedule, the bills of material file, and the
inventory records file. Product demand data for MRP systems come from two sources – from the
customers who have placed firm orders and from forecasted ones. The master schedule will level
Downloaded by FEI At 18:06 07 January 2018 (PT)

the load, so the requirements for work centers remain within the available capacity and inventory
policy. Purchase and production orders are automatically sent to the suppliers and programmed
for production, respectively, when they reach minimum levels of inventory for finished product.
As the demand forecast re-feeds the beginning of the process instead of the effective
demand, it corresponds to the concept of a push system. The tendency is that the warehouse
and intermediate inventory are maintained elevated, even during periods without demand.
The MRP did not efficiently correspond to the company’s needs regarding three
particularities of the scenario:
(1) the variable mix of products tends to elevate inventory levels (storage at all levels);
(2) the MRP concept is complex and depends on reliable and non-existing statistical
information, causing an erroneous programming of production; and
(3) the lack of demand predictability, added to erroneous programming, resulted in the
customers’ needs not being met, generating long delivery times and a low level of service.
The company’s general manager declared that a lack of coordination and harmonization
with MRP leads to inefficient, competitive, and even contradictory interventions with
negative results. The next attempt to improve process was via Kanban. This methodology
was implemented in early 2012, conceiving its production line as illustrated in Figure 8.
In order to adapt production with the Kanban control method, we identified the good
production batches for each piece and calculated them, i.e. the number of Kanban cards, as well
as the levels of WIP at each stage of production: store, supermarket, line, and finished product.

Purchasing Production Production Production

order order order order

Suppliers
Figure 7.
Raw
Production Production Production Forecast MRP production
material
control
Source: The Company
 JMTM When the inventory of finished products reaches the minimum pre-determined level,
28,6 the Kanban card is transferred to the line, which is its previous inventory. This can also
happen in case of sales orders. This process is repeated continuously among previous
stages, supermarket and store, until the balances are sufficient to meet the demand.
This corresponds to the concept of a pull system on account of the loading process only
occurring in response to a demand, and the tendency is that inventories always maintain half
724 of the stored demand, even without demand. When a process stops, the following inventory’s
process allows it and the others to proceed, at least until the content of that buffer ends.
The implementation of Kanban still did not efficiently correspond to the needs of the
company studied regarding three particularities:
(1) The variable mix of products tends to raise inventory levels at the store,
supermarket, and line because the demand for each item is very unstable.
(2) The varied demands prevent information from arriving rapidly to the beginning of
the chain, delaying the release of purchase orders, and consequently the non-
fulfillment of established delivery times.
(3) The same production operators execute the movement processes, maintaining the
Downloaded by FEI At 18:06 07 January 2018 (PT)

inventory effect neutralized. This happens because they have to stop production in
order for the movement to happen, hindering the efficiency of this method, and
losing the essence of Kanban.
In search for a method, which explained the reality of the company, we implemented
production control via CONWIP. In early 2013, its production line was conceived, as
illustrated in Figure 9.
The CONWIP control method can be considered a single-stage Kanban. The stand-by
intermediate inventory levels are null, and only the inventory levels for the finished product
are established.
When the inventory level for finished products reaches its minimum level, and in case of
sales order, a production order is generated for the first stage, causing the pieces to be
pushed as fast as possible through the system.

Purchasing
order Kanban Kanban Kanban Kanban

Figure 8. Store Supermarket Line Sales order

Supplier Production Production Production
Kanban system
Source: The Company

Purchasing order
Production order

Sales order
Figure 9. Store Supermarket Line Final product
CONWIP Supplier Production Production Production
production control
Source: The Company
 This corresponds to the concept of a pull system due to the loading process only occurring Hybrid
in response to a demand. The store’s inventory, supermarket, and line are exhausted Kanban-
because they are dimensioned according to the batch of sales order. The feasible operations CONWIP
are advanced until the compromised pieces are exhausted.
Its main disadvantage is a tendency to accumulate pieces at the buffer before the system
bottleneck, or where the process stopped. Besides, as the concept was designed for
productive instead of loading processes, it becomes illogical when applied to suppliers, since 725
the delivery time of the inputs is long and cannot meet the real demand.
Therefore, the re-supply of external suppliers needed to be treated separately through
demand forecast statistical calculations, based on the records, and it deals with storage at
suppliers due to its long delivery time and multiple needs, as Figure 10 shows.
The CONWIP control method was the one that better adapted to the scenario of the
company studied until 2014. However, it still did not fully correspond to the needs of the
market regarding some particularities:
• This method greatly minimizes the levels of inventory in line when compared to
Kanban because a component is only moved when demand really occurs to the final
item. However, the variable mix of products is still responsible for considerable levels
Downloaded by FEI At 18:06 07 January 2018 (PT)

of inventory in other places.

• The production of small batches, associated with the variable mix of products, makes
it difficult to meet the delivery times jointly defined with the clients.
Therefore, the CONWIP method presents better results when compared to Kanban, but not
enough to meet the needs of the market.
An important particularity was met:
• the delay effect caused by varied demands was reduced to its minimum as the
information directly arrives at the beginning of the chain, pushing the production
more rapidly.
While looking for academic researches, the operational manager found a study from MIT on
the internet, carried out by Bonvik et al. (1997), which optimizes production and delivers a
better customer service performance. The manager then found a similarity between the
study and the reality of his production line (composed of up to four assembly stages)
regarding the variable mix of products and varied, unpredictable demands. Analyzing the
propositions and benefits presented in this study, the operational manager concluded that
the application of the hybrid CONWIP/Kanban to control production would bring better
results, being implemented in early 2014.
This new methodology conceives the scenario illustrated in Figure 11.
Similarly to CONWIP, the re-supply of external suppliers needs to be treated separately
via MRP due to its long delivery time and multiple needs.

Purchasing order
t1
Production order

Sales order

Store Supermarket Line Final product Figure 10.

Supplier Production Production Production CONWIP production
control with suppliers
Source: The Company
 JMTM The levels of stand-by intermediate inventories are basically null, and only the inventory
28,6 levels for the finished product are established.
When the inventory of finished products is empty and sales orders are present,
a production order is generated for the first stage of CONWIP. As the finished product
demand feeds the beginning of the chain, it corresponds to the concept of a pull system.
We could observe that a large part of the marketed products, despite their high level of
726 customization, had a standardized level until a certain stage (sub-groups). With this
new method, before the production order went to production, the system generated a list of
sub-groups, which would be used in the products to meet the sales order, grouping and
sending them to production through Kanban. Therefore, the semi-finished production was
controlled through Kanban, which determined good batches from each sub-group.
The pieces are pushed as fast as possible through the system until they arrive at
the final production, which used the sub-group to customize the finished product to meet
the sales order.
In this scenario, the hybrid CONWIP/Kanban completely met the needs of the company
regarding three particularities:
(1) The variable mix of products tends to raise inventory levels. On the other hand, with
Downloaded by FEI At 18:06 07 January 2018 (PT)

the increase in production rate, the products had a faster turnover, reducing the
inventory levels even more than in CONWIP. The increase in inventory caused by
the product mix is compensated by the speed of the process.
(2) Even with the variable mix of products and low volume production, the delivery
times were met through the agility that the production of sub-groups in
Kanban offered to the system, generating better levels of service than the ones
previously reached.
(3) Similarly to CONWIP, the delay effect caused by varied demands was basically
neutralized because the information arrives directly to the beginning of the chain,
pushing the production more rapidly.
Throughout every stage of the problem-solving process, we could observe the behavior of
the total inventory level ( finished product, inventory in process – work-in-progress, and
raw material).
The professionals involved with this project concluded that the inventory level decreased
as the method was more adapted to the process. Figure 12 aims at illustrating this statement.

4.5 The analysis of the production system at a product level

In spite of the huge number of product families, a set of characteristics are common to all of
them, determining the conception of the production area layout.

Production order

Production order

Sales order

Figure 11. Store Semi-finished good Semi-finished good Final product

Inventory Supplier Production Production Production
level behavior
Source: The Company
 Hybrid
Kanban-
MRP
CONWIP
system
KANBAN
Inventory level

727
CONWIP

CONWIP WITH
SUPPLIERS

CONWIP/KANBAN Figure 12.

HYBRID
CONWIP/KANBAN
hybrid production
Time
control
Source: The Company
Downloaded by FEI At 18:06 07 January 2018 (PT)

Basically, almost all the products are composed of a kind of box (in different materials and
designs) with the respective cover, some kind of a visual indicator on the top and a bracket on the
bottom (also with a large number of designs), to couple the unit to hundreds of different valves.
Inside the box, several kinds of electronic or electromechanical devices can be
accommodated, depending upon the specifics of every valve, on a one by one basis. So,
although there are hundreds of possible combinations, we can say that there are important
characteristics that are common to all of them:
• They all have a housing (with dimensions and weight approximately similar and can
be handled unitarily by a person without the need of special tools).
• The quality requirements in terms of final product tests and inspections may be
divided in a small number of categories.
• The products are mostly assembled in production lines from buyout components.
Very simple and light machinery is necessary.
Based mainly on these characteristics and on the expected capacity, the plants were
designed the way they are:
• four assembly lines with four working stations each;
• four single position cells, for special tooling production;
• reserved space on both extremes of every line and corridor space for cart traffic along
the line, opposite to the operator’s side; and
• the corridors between the shelves of the stockroom must be wide enough for the carts
to transit along them.
Figure 13 shows the layout of the company under study in its final configuration, with the
hybrid CONWIP/Kanban system implemented.
The company’s production process is divided into two stages – the sub-assembly and the
final assembly processes. Simply put, the productive process can be described as an initial
stage, which encompasses the arrival, storage, and preparation of component batches
to supply the sub-assembly area. This process carries out the assembly of the central part of
the valve command and the final assembly, which integrates the final components
customized by the end customer. Figure 14 illustrates the main stages of this process.
 JMTM D.T. D.T.
Depo Development
Manager
28,6 Vehicles load/unload Future growth
engineering

Purchase and
logistics
engineering
Meeting

Metrology
Product
room P.C.P.

Reception
728 Quality

Sepa
control

Stockroom Separation

Line 1

Line 2

Line 3

Line 4

Leasure area
Shipping

Production
Packing
Figure 13.
Cell 4 Cell 3 Cell 2 Cell 1
Layout
Downloaded by FEI At 18:06 07 January 2018 (PT)

Inventory
Batch Sub- Final
Packaging
preparation assembly assembly

Figure 14.
Stages of the
productive process

The batch preparation stage is responsible for the separation of all components, which will
be used at the sub-assembly and/or final assembly processes. It refers to a group of
employees in charge of this activity – technicians with expertise in identifying the correct
component for each product to be produced. These parts are only sent to the assembly lines
when needed. After receiving the components, the sub-assembly line carries out the basic
manufacturing of the valve command body, and then sends the assembled set to a
supermarket of semi-finished products. After receiving the order from the final customer,
the batch preparation area once again separates the components required by the end
customer and the previously pre-assembled sub-set, and then sends these materials to the
final assembly area, which carries out the assembly desired by the customer, forwarding
the final product to packaging. Finally, they are shipped to the end customer.
The only exception to this sequence are the high-volume products, which are completely
assembled without passing through the supermarket of semi-finished products. As a
sample, Table III shows the results of this process in the production line of a Group B family.
First, the company decided to implement a production control through MRP, and as
Table III shows, after a maturation period the company’s results were: inventory level was
at 239 pieces, utilization was 50 percent and time spent on one product was 60 days.
The variable mix of products tends to elevate inventory levels. The worst record in the
Table III is order delivery reliability to customer, which remained only on 50 percent of all
orders of this specific product family.
 Downloaded by FEI At 18:06 07 January 2018 (PT)

“B” Family with demand ¼ 100 uni./month

Production stages MRP Kanban CONWIP CONWIP/Kanban hybrid
Inventory Utilization (%) Time/Piece Inventory Utilization (%) Time/oiece Inventory utilization (%) Time/oiece Inventory utilization (%) Time/oiece

(1) Batch preparation 0 0 30 40 10 80 6 90

(2) Sub-assembly 80 40 55 52 12 77 8 88
(3) Semi-finished goods 60 days 55 days 20 70 25 days 15 92 20 days
(4) Final assembly 94 45 46 43 8 72 4 90
(5) Package 65 48 55 45 7 65 2 88
Finished goods 50 40 80 93
WIP 239 186 57 35
Delivery reliability 50 40 80 93
Customer order 100 100 100 100
Kanban-
Hybrid

729
system

Results at a Group B
CONWIP

family level
Table III.
 JMTM The next attempt to improve process was via Kanban. Kanban is a pull system that means
28,6 the customer orders are already known in advance and are not produced to stock as it is in
push system but they are made to order. The results are shown in Table III. From the data,
it is seen that the WIP fell from 239 to 189 pieces. The variable mix of products tends to
raise inventory levels. Utilization was 40 percent and the time needed to produce one
product is 55 days.
730 In search for a better method, the company decided to implemented production control
via CONWIP. The CONWIP control method can be considered a single-stage Kanban.
The stand-by intermediate inventory levels are null, and only the inventory levels for the
finished product are established. When the inventory level for finished products reaches its
minimum level, and in case of sales order, a production order is generated for the first stage,
causing the pieces to be pushed as fast as possible through the system. The results are
shown in Table I. From the data, it is seen that the WIP fell from 186 pieces to 57 pieces. This
method greatly minimizes the levels of inventory in line when compared to Kanban because
a component is only moved when demand really occurs to the final item. Utilization was
80 percent and the time needed to produce one product is 25 days.
As a final planning system, the company used hybrid CONWIP/Kanban to control
Downloaded by FEI At 18:06 07 January 2018 (PT)

production. Similarly, to CONWIP, the re-supply of external suppliers needs to be treated

separately via MRP due to its long delivery time and multiple needs. Production results are
shown in Table III. From the data, it is seen that the stock is 35 pieces, utilization is
93 percent and the time required to produce one product is 20 days. We must not forget also
on the ability to fulfill orders the in this case, it is at 93 percent success rate. Of all the
periods of production in which the company worked, has nothing changed in the processes,
only the method of planning.
Improving the reliability and throughput using the hybrid CONWIP/Kanban system lies
in the fact that manufacturing orders are pushed to the production processes, but only to
the extent that the production volume reaches its maximum capacity. After the release of the
finished order, other commands can be pushed to the production. New orders can be pushed
to the production not only when there is enough space, but also after a signal, or after
crossing a minimal state level of the stock.
An important indicator of a company’s operating performance is inventory turnover (IT).
In the case of this type of enterprise, this issue is quite critical because with the large
variety of items with low unitary consumption, IT tends to be very low. Although high
levels of turnover cannot be achieved anyway, Table III denotes the strong benefit impact
that the WIP implementation brought to the company. Considering that around 2014 the
company had $20 million dollars, of revenue, this measure allowed a reduction of half of
this amount in working capital immobilization and a 40 percent reduction in the area
occupied by inventories.
The use of the hybrid CONWIP/Kanban planning system in this medium-sized company
was very helpful to plan and control its production, as it kept its inventory values low,
presented more efficient work, and optimized the process with low costs. It provided the
customer with the best quality for the best price.

4.6 The analysis of the production system at aggregate level

In order to express the results obtained, we have considered only the three main years,
period in which the production control methods were implemented, as Table IV shows.
The first result we will present is the on-time delivery (OTD), as shown in Figure 15.
OTD had a significant evolution when compared to different years, as analyzed in the
graph of Figure 15.
In 2012, OTD presented an average of 53 percent with the use of the Kanban method;
however, the average in 2013 was 72 percent with the CONWIP method. Finally, with the
 use of the hybrid CONWIP/Kanban, the service was elevated to 89 percent. It improved by Hybrid
34 percent between 2012 and 2013, and by 24 percent between 2013 and 2014. Kanban-
Figure 16 identifies the evolution of delivery time continuously throughout the three CONWIP
years analyzed to better verify this growth.
By analyzing the graph of Figure 16, we can verify that OTD had a growth rate of system
approximately 30 percent per year during the three-year period, and 1.2 times at the end of
this period. This demonstrates that this improvement is connected to the evolution of the 731
company’s production control methods.
With the implementation of control via the hybrid CONWIP/Kanban in 2014, the
company was able to meet the established goal for OTD in most months, reaching levels
above 90 percent.
Another significant improvement was the IT, which can be verified in Figure 17.
The IT had its worst result during the period that MRP controlled production (2007-2014),
reaching an average turnover of only 1.8 times per year due to the high inventory level needed
to meet it.
With the maturity of MRP, the IT increased to 2.6 times per year at the end of its use.
During the following three years, the IT increased from 3.2 to 3.4 and 3.6 times per year
Downloaded by FEI At 18:06 07 January 2018 (PT)

with the implementation of Kanban, CONWIP, and the hybrid CONWIP/Kanban.

Other important gains of the company, whose data are not illustrated in specific figures,
were: an increase of 40 percent in productivity without the increase of headcount, when
comparing the years of 2010 and 2014; an increase of 18 percent in revenue during the same
period; and a reduction of more than 70 percent in slow moving items due to the high
rationalization of inventories, which were reduced by 20 percent of their value and almost 50
percent of their area.
During the interview with the operational manager, we collected and classified factors
related to the implementation of the hybrid CONWIP/Kanban, according to Figure 18.

Year Production control method

2012 Kanban Table IV.

2013 CONWIP Period × production
2014 CONWIP/Kanban hybrid control

100
90
80
On-time delivery %

70
60
50
40
30
20
10
0
Out Nov Dez Jan Fev Mar Abr Mai Jun Jul Ago Set
2012 60 41 34 58 46 62 61 57 64 57 44 56 Figure 15.
2013 53 32 74 71 77 73 76 80 83 83 81 78 On-time delivery –
2014 75 84 91 90 92 88 89 93 91 92 91 92 annual from October
2012-September /2014
Source: The Company
 JMTM 2012 2013 2014
28,6 100

90

80

70
On-time delivery %

732 60

50

40

30

20

10
Figure 16.
On-time delivery – 0
O N D J F M A M J J A S O N D J F M A M J J A S O N D J F M A M J J A S
linear from October
Downloaded by FEI At 18:06 07 January 2018 (PT)

OTD 60 41 34 58 46 62 61 57 64 57 44 56 53 32 74 71 77 73 76 80 83 83 81 78 75 84 91 90 92 88 89 93 91 92 91 92
2012-September 2014
Source: The Company

12.0 4.0

3.5
10.0
3.0

Inventory turnover
Million dollars

8.0
2.5

6.0 2.0

1.5
4.0
1.0
2.0
0.5

0.0 0.0
2007 2008 2009 2010 2011 2012 2013 2014
Inventory 679 1,189 3,141 2,905 2,471 2,670 2,710 2,800
Figure 17. Sales 2,070 2,475 3,870 5,229 6,425 8,544 9,214 10,080
Evolution-inventory Inv. Turn. 3.1 2.1 1.2 1.8 2.6 3.2 3.4 3.6
turnover
Source: The Company

Observing the results from Figure 18, we can verify that the support from the directors,
the choice of a multifunctional team, the user’s involvement, the choice of a leader for the
project, and company culture were decisive factors for the success of the implementation of
the hybrid CONWIP/Kanban.
No external consultant was involved during the development of the project, as the
method at issue was little explored, especially here in Brazil.
It is important to highlight the lack of a defined methodology to implement the method,
which reinforces the project leader’s acumen and his capacity to involve and mobilize the
team. As observed, the suppliers’ support was essential to understand the problem and
work on its solution.
We can perform a synthetic analysis of Figure 19 regarding the results obtained from
this implementation by assessing some parameters developed throughout this study.
Among them, we can highlight: an expressive reduction in forecast errors, i.e. programming
 Was a methodology used to implement the method? Hybrid
Were external consultants used? Kanban-
Were there minimal customizations?
Were the project’s objectives clear and shared? CONWIP
Was the company culture helpful? system
Did the company’s IT department offer assistance?
Did the project have a leader?
Did the software perform as expected? 733
Was the supplier’s involvement adequate?
Was the user involved?
Was the team trained?
Did all involved areas collaborate?
Was the implementation team functional?
Is the CONWIP system compatible with ERP? Figure 18.
Was the project monitored and evaluated? Results obtained by
Did the directors support it?
implementing the
hybrid CONWIP/
0 1 2 3 4 5 Kanban
Strongly disagree Strongly agree
Downloaded by FEI At 18:06 07 January 2018 (PT)

Mix of products adequate to the available…

Improvement of quality and reduction of…
Reduction in the amount of workforce in…
Reduction in manufacture lifecycle
Increase in planning velocity
More stable and errorless IT systems
Reduction in delivery time
Increase in the level of customer service
Reduction of activities that do not…
Production more synchronized with the…
Reduction in levels of Inventory
Reduction of forecast errors
Figure 19.
Cost reduction Factors that affect the
0 1 2 3 4 5 hybrid CONWIP/
Kanban
Strongly disagree Strongly agree

of the production adapted to the demand; decrease in levels of inventory; improvement in

delivery times; reduction in workforce in process; and especially increase in the level of
customer service and improvement in the quality and reduction of product defects due to
wrong assembly or change of components during production.

5. Conclusion, managerial implications, limitations, and future research steps

This study aimed at understanding, describing, and investigating the systems for
coordinating purchase and production orders regarding Kanban, CONWIP, and the hybrid
CONWIP/Kanban, implemented in a real production environment, in order to identify their
results and main difficulties found. According to Lage Junior and Godinho Filho (2008):
[…] the study of application in practice to adapt KANBAN systems only theoretically developed is
a theme with several opportunities for exploration.
 JMTM This was the objective of this work.
28,6 The study investigated the system implemented in a manufacturer of hi-tech
electromechanic products, including innovative solutions to command and monitor valves.
Their operations executives tried several systems for coordinating purchase and production
orders from 2010 to 2014, including MRP, Kanban, CONWIP, and the hybrid CONWIP/Kanban
as Bonvik et al. (1997) proposed, focusing on meeting their demand goals with the lowest cost.
734 The theoretical and practical learning allowed the company’s executives to understand
their processes and obtain positive results. The small-batched productive system with high
customization and uncertain demand became more efficient with the implementation of the
hybrid CONWIP/Kanban, as the summary the company’s main indicators shows in Table V.
These results indicate that the operational performance with the hybrid CONWIP/
Kanban was superior to the pure systems tested. Therefore, we can affirm that this
company aims at searching for hybrid systems, mainly due to the increase in variety of
products that the market demands, followed by a reduction of the volumes requested. This
demands more flexibility and velocity from the productive system.
In addition to the reports on the results found, we investigated some academic research
questions, and here are their answers:
Downloaded by FEI At 18:06 07 January 2018 (PT)

RQ1. Which variables have an impact on selecting the system for coordinating purchase
and production orders, aiming at maximum level of service with lowest cost?
This company’s peculiar characteristic indicate that the following characteristics are
relevant and influence decision making when selecting the system for coordinating
purchase and production orders: volume of production (high or low), level of demand
predictability (high or low), size of production batches (high or low), level of repetition,
diversification of the product mix regarding the families of products, and distinction
between each family regarding the variety of models:
RQ2. When is the Hybrid CONWIP/Kanban more appropriate in productive systems?
Table VI shows a summary of the characteristics the researched company’s executives find
to be important.
The hybrid CONWIP/Kanban is the most efficient one for productive systems dedicated
to small batches and high level of customization:
RQ3. What are the results obtained with the application of the hybrid CONWIP/Kanban
in a company of customized products, small batches, and uncertain demand?

CONWIP/
Production aspects Kanban Kanban/MRP hybrid CONWIP CONWIP/MRP hybrid Kanban hybrid

Demand Stable Predictable Unstable Predictable Unstable

Product variability Low Low High High High
Table V. Volume High High Low Low Low
Production aspects vs Inventory level High High high Moderate Low
control methods Lead time Just in time Just in time High Moderate Low

CONWIP/MRP
Results Kanban Kanban/MRP hybrid CONWIP hybrid CONWIP/Kanban hybrid
Table VI.
Results obtained vs Level of service Low Low Moderate Moderate High
control methods Inventory High High Moderate High Low
 The lessons learned in this case studied are very important, especially because they refer to Hybrid
a Brazilian company that is presenting positive and concrete results with the adoption of the Kanban-
hybrid CONWIP/Kanban. The executives’ testimonials revealed that the Hybrid CONWIP/ CONWIP
Kanban is not a system for all types of organizations; we need to consider a series of
environmental and marketing conditions, as well as operational requirements, whose system
presence is mandatory to enable its adoption. The question is not wanting to implement the
hybrid CONWIP/Kanban, but being able to do so. The organization needs to have minimum 735
conditions of structure, infrastructure, and inter-organizational relationship for its
implementation. Another important point is the cultural aspect involved in the
implementation. Not only does the system’s efficiency depend on the present technical
operational conditions, but it also depends on the transformation of relations among sectors
of the company and other companies. In this context, it is important to break organizational
barriers that usually hinder collaborative relationships. As Fisher (1997) states, the project
of a responsive process is not a simple task; however, the reward – a notable competitive
advantage – can make these efforts worth it.
It is important to mention some research limitations. First, the impossibility of statistic
generalization of the results due to the use of a single case method of study. Second,
Downloaded by FEI At 18:06 07 January 2018 (PT)

the research evaluated the results obtained wraith several systems for coordinating
purchase and production orders; however, it had the disadvantage of studying only one
case from each category.

5.1 Future research works

We recommend some points for future studies:
Research on different sectors, such as electronics, represents a contribution to evaluate the
process of spreading the practices of the hybrid CONWIP/Kanban in the Brazilian market.
Several authors have studied the impact of pull programming on a company; however,
few of them have investigated the impact on the members upstream or downstream the
supply chain. A relevant study would be to investigate this theme.
More focused reproductions of this study in the hybrid CONWIP/Kanban would bring
additional contributions.

References
Bonvik, A.M., Couch, C.E. and Gershwin, S.B. (1997), “A comparison of production-line control
mechanisms”, International Journal of Production Research, Vol. 35 No. 3, pp. 789-804.
Boonlertvanich, K. (2005), “Extended-CONWIP-Kanban system: control and performance analysis”
Doctoral dissertation, Georgia Institute of Technology.
Eisenhardt, K.M. (1989), “Building theories from case study research”, Academy of Management
Review, Vol. 14 No. 4, pp. 532-550, doi: 10.5465/AMR.1989.4308385.
Ellram, L.M. (1996), “The use of the case study method in logistics research”, Journal of Business
Logistics, Vol. 17 No. 2, p. 93.
Farnoush, A. and Wiktorsson, M. (2013), “POLCA and CONWIP performance in a divergent production
line: an automotive case study”, Journal of Management Control, Vol. 24 No. 2, pp. 159-186.
Fernandes, F.C.F. and Godinho Filho, M. (2010), Planejamento e controle da produção: dos fundamentos
ao essencial, Atlas, São Paulo.
Fisher, M.L. (1997), “What is the right supply chain for your product?”, Harvard Business Review,
Vol. 75 No. 2, pp. 105-116.
Framinan, J.M., González, P.L. and Ruiz-Usano, R. (2003), “The CONWIP production control system:
review and research issues”, Production Planning & Control, Vol. 14 No. 3, pp. 255-265,
doi: 10.1080/0953728031000102595.
 JMTM Gaury, E.G.A., Pierreval, H. and Kleijnen, J.P.C. (2000), “An evolutionary approach to select a pull
28,6 system among Kanban, CONWIP and hybrid”, Journal of Intelligent Manufacturing, Vol. 11
No. 2, pp. 157-167.
Gilland, W.G. (2002), “A simulation study comparing performance of CONWIP and bottleneck-based
release rules”, Production Planning & Control, Vol. 13 No. 2, pp. 211-219.
Huang, M., Wang, D. and Ip, W.H. (1998), “Simulation study of CONWIP for a cold rolling plant”,
International Journal of Production Economics, Vol. 54 No. 3, pp. 257-266.
736
Lage, M. Jr and Godinho Filho, M. (2008), “Adaptations of the Kanban system: review, classification,
analysis and evaluation”, Gestão & Produção, Vol. 15 No. 1, pp. 173-188.
MacCarthy, B.L. and Fernandes, F.C.F. (2000), “A multi-dimensional classification of production
systems for the design and selection of production planning and control systems”, Production
Planning & Control, Vol. 11 No. 5, pp. 481-496.
Muckstadt, J.A. and Tayur, S.R. (1995), “A comparison of alternative Kanban control mechanisms.
I. Background and structural results”, IIE Transactions, Vol. 27 No. 2, pp. 140-150.
Ni, Y. and Wang, Y. (2015), “A double decoupling postponement approach for integrated mixed flow
production systems”, Kybernetes, Vol. 44 No. 5, pp. 705-720.
Ohno, T. (1988), Toyota Production System: Beyond Large-Scale Production, CRC Press.
Downloaded by FEI At 18:06 07 January 2018 (PT)

Prakash, J. and Chin, J.F. (2014), “Implementation of hybrid parallel Kanban-CONWIP system: a case
study”, Cogent Engineering, Vol. 1 No. 1.
Prakash, J. and Chin, J.F. (2015), “Modified CONWIP systems: a review and classification”, Production
Planning & Control, Vol. 26 No. 4, pp. 296-307.
Sharma, S. and Agrawal, N. (2009), “Selection of a pull production control policy under different
demand situations for a manufacturing system by AHP-algorithm”, Computers & Operations
Research, Vol. 36 No. 5, pp. 1622-1632.
Shingo, S. and Dillon, A.P. (1989), A Study of the Toyota Production System: From an Industrial
Engineering Viewpoint, Productivity Press, New York, NY.
Slomp, J., Bokhorst, J.A.C. and Germs, R. (2009), “A lean production control system for high-variety/
low-volume environments: a case study implementation”, Production Planning and Control,
Vol. 20 No. 7, pp. 586-595.
Spearman, M.L., Woodruff, D.L. and Hopp, W.J. (1990), “CONWIP: a pull alternative to Kanban”,
The International Journal of Production Research, Vol. 28 No. 5, pp. 879-894.
Yang, T., Fu, H.-P. and Yang, K.-Y. (2007), “An evolutionary-simulation approach for the optimization
of multi-constant work-in-process strategy – a case study”, International Journal of Production
Economics, Vol. 107 No. 1, pp. 104-114.
Yin, R.K. (2013), Case Study Research: Design and Methods, Sage Publications, London.

Corresponding author
Mauro Sampaio can be contacted at: msampaio@[Link]

For instructions on how to order reprints of this article, please visit our website:
[Link]/licensing/[Link]
Or contact us for further details: permissions@[Link]

View publication stats